Did the sub-gigahertz G4 win out over the 2.0GHz Intel behemoth? Well, yes and no. In terms of win-loss, the Pentium 4 handily beats the G4 scoring five wins to one. However, the G4 showed impressive strength in the RGB-to-CMYK color-conversion test by turning in a time of 23.6 seconds compared to 38.8 seconds on the P4 system -- a noticeable 64 percent quicker (view results). This particular test was the black eye that brought down the P4's overall average.

Image rotation scored the Pentium test box a small victory. The P4 was able to finish this task in 16.3 seconds compared the latest G4 at 17.5 seconds -- 7 percent faster.

The P4 won the Gaussian Blur test and Unsharp Mask test. The G4 managed to turn in times of 9.1 and 11.0 seconds respectively, but the P4 bested it with times of 6.7 and 4.7 seconds -- 36 percent and 134 percent faster, respectively (see results). Impressive percentages, but in terms of actual time the differences weren't enough to average out the P4's color-conversion defeat.

Image reduction and Lighting Effects were our two final tests. The P4 was able to finish reducing the image in 1.7 seconds -- about 88 percent faster than the G4's time of 3.2 seconds (see results). The P4, with its optimized Lighting Effects plug-in, managed to win this test with a time of 8.6 seconds beating the G4 and its time of 9.8 -- averaging 14 percent faster (view results).

Final thoughts

The overall times recorded to complete our six Photoshop tests favored the Mac G4 by a slim three percent. If we ignored the RGB to CMYK conversion test results, the Pentium 4 would have swept this competition and left the G4 behind by a healthy 34 percent.

The dark side of benchmarking computers is that anyone can put together a scenario in which one computer is shown to dominate another -- aka selective benchmarking. Only with well-written and equally optimized applications for differing platforms can one hope to come away with any meaningful results.

This particular benchmark focused on photo editing. Tests like these are not the best representation of overall performance. Nor do they accurately depict the end-user's experience. You can be sure that this topic will be constantly revisited and updated by TechTV Labs.